# Agentic GraphRAG

To learn the basics of GraphRAG, take a look under the [Atomic GraphRAG
Pipelines](https://memgraph.com/docs/ai-ecosystem/graph-rag/atomic-pipelines).

The combination of Atomic GraphRAG Pipelines and Agents adds another dimension
the the whole story because an agent can on it's own figure out the right set of
primitives to get the job done (answer a question or perform a task).

Using the [GraphRAG
Skill](https://github.com/memgraph/skills/tree/main/skills/memgraph-graph-rag), it's
possible to guide and agent to start using appropriate GraphRAG Pipelines based
on the user question/prompt. For more skills, visit [Memgraph Agent
Skills](https://github.com/memgraph/skills). Also, make sure you have the
[MCP](https://memgraph.com/docs/ai-ecosystem/mcp) properly configured (connected to the Memgraph
database), because any given agent will primarly use [MCP](https://memgraph.com/docs/ai-ecosystem/mcp) to
executed the queries. If you use, [Memgraph Lab](https://memgraph.com/docs/memgraph-lab) as the agentic
runtime / agent, everything should already be in-place.

An example of a prompt under [Memgraph Lab GraphChat](https://memgraph.com/docs/memgraph-lab/features/graphchat) follows:
```
Using https://github.com/memgraph/skills/blob/main/memgraph-graph-rag/SKILL.md,
what's the most important person in dataset?
```

Under the hood, you will end up with something like below image:
![agentic_graphrag_pipelines](https://memgraph.com/docs/pages/ai-ecosystem/graph-rag/agentic-graphrag/agentic-graphrag.png)